Concurrency aware artificial intelligence for real-time intrusion detection in software-defined networks
Year: 2026
Type: article
Abstract: Contemporary software systems are now being used in environments that are highly concurrent, data-intensive, and latency-sensitive. Software-Defined Networks (SDNs) applications like real-time intrusion detection need fast pattern recognition in large amounts of data streams through time. Although deep learning models have impressive detection capabilities, failure to perform well in real-world deployments is usually limited by poor concurrency management, inflexible scheduling, centralized learning pipelines, and fault tolerance. In this study, the authors suggest a concurrency-aware artificial intelligence (CA-AI) framework that directly incorporates actor-based concurrency control, workflow-based load-aware scheduling, deep long short-term memory (LSTM)-based temporal learning, agent-based parallel feature extraction, and feedback loops based on monitoring. The framework is evaluated for SDN-oriented intrusion detection workloads, where concurrency management is implemented at the system architecture level through actor-based execution, workflow scheduling, and asynchronous inference and tested on real network traffic data samples (NSL-KDD, CIC-IDS2017, and UNSW-NB15). Experimental results demonstrate that concurrency-aware system design leads to a significant improvement in the detection accuracy achieving up to 98.9% detection accuracy across multiple benchmark datasets, including NSL-KDD, CIC-IDS2017, and UNSW-NB15, throughput improved by 2.1 × compared with the centralized IDS baseline, while throughput increased by approximately 1.19 × when concurrency increased from 10,000 to 100,000 flows within the proposed CA-AI framework, average latency was reduced by approximately 57.5% compared with the centralized IDS baseline (146 ms to 62 ms), while maintaining high reliability under concurrent workloads. The proposed framework establishes concurrency as a fundamental architectural principle for scalable AI-driven pattern recognition systems.
Source: Scientific Reports
Authors: R. Kanimozhi, V. Padmavathi, M. Latha
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7223092664